Feature
Your AI Team for the Customer Lifecycle
Deploy specialist agents for pre-sales, lead qualification, closing, and customer success. Coordinate them in groups, share the relevant customer context, and keep people in control of important actions.
Capabilities
Specialists that act, hand off, and improve
Chat & Email Agents
Embed a chat widget on your website for pre-sales, lead qualification, or customer support. Connect email channels for automated triage. Each agent gets its own personality, prompt, and channel configuration.
Multi-Agent Routing
A triage agent reads incoming messages and routes them to the right specialist — pre-sales, support, or billing. Agents hand off to each other with full conversation context preserved.
Human-in-the-Loop
Review agent decisions before they go out. Approve, edit, or reject responses, then use corrections as guidance for similar situations.
AI Tool Use
Agents dynamically choose which actions to take — look up a contact, update a deal, send an email — based on conversation context. They reason about what to do, not just follow a script.
Correction Memory
When you correct an agent, Outrun can retrieve relevant past corrections as examples for similar situations. Keep reviewing outcomes as the workflow changes.
Visitor Context
Use the SDK to identify visitors on your site. Agents see who they’re talking to — company, role, past interactions — and tailor responses accordingly.
Agent template library
Start with a customer-facing role or an operational assistant, then combine specialists in an agent group when the workflow needs a handoff.
Customer Lifecycle
Prospecting & Follow-up
Account Operations
Recurring Work
Assistants
Service & Data
Your data is already connected. Just add AI.
Connect the records, conversations, and knowledge a role needs. Agents can retrieve context or run supported actions through those connections; exact capabilities vary by integration.
How agents work with your tools
Website Chat
Embed a chat widget powered by AI agents. Visitors get instant, context-aware responses. Agents identify visitors and tailor conversations using your CRM data.
Agents read, classify, and respond to email. They extract data, update your CRM, and escalate when they’re not confident.
CRM & Support
Agents look up contacts, update deals, create tickets, and pull context from HubSpot, Pipedrive, Zoho, Zendesk, and more.
Agents that improve
Correction memory in practice
First correction
Your agent drafts a follow-up email that’s too formal for your brand. You edit it to match your tone and approve the revision.
Similar situation
When a similar prospect arrives, the agent can retrieve the correction as an example and use it when preparing the next draft.
Ongoing review
Review the outcome, keep useful guidance, and continue requiring approval where the decision is sensitive or the situation is new.
Operating model
Start narrow. Earn trust. Expand deliberately.
A useful agent is more than a prompt. Give each role a clear job, the minimum context and tools it needs, and an explicit approval posture.
Define the role
Choose one measurable outcome and document what the agent should never do on its own.
Review the work
Inspect activity, proposed actions, and handoffs. Correct the instruction or context behind a bad result.
Add the next specialist
Create an agent group only after the first role is reliable enough to hand work to another specialist.
Ready to get started?
Choose one role, connect the context it needs, and validate the workflow with human approval.
Review current plans and allowances on the pricing page
